Cancel
Showing results for 
Search instead for 
Did you mean: 

Femap DB size increase w/ attached XDB files

Creator
Creator

We have a large aircraft internal loads model with hundreds of load conditions, each in a separate XDB. Although it states the results aren't internalized, I've noticed a pretty major impact to the size of the model file when attaching a large number of results files. For example, a 72MB DB without results attached has grown to 1GB with 120 results files attached. The database will be prohibitively large with all of our conditions attached, so we're forced to use another software package for our internal loads model processing.

 

Is there a way to prevent this, or to at least compact the database after attaching results? I tried a rebuild, but that didn't have any effect.

 

Thank you

7 REPLIES

Re: Femap DB size increase w/ attached XDB files

Siemens Phenom Siemens Phenom
Siemens Phenom

Can you tell us a bit more about your Results?

  • With your 120 XDB Files attached, how many Result Sets does FEMAP show?
  • About how large are your XDB files? either individually or in total ?
  • How many total files/Result Sets do you need to attach?

While it is true that the Results are not stored in the FEMAP database when you attach a file, there is some information stored so that FEMAP can quickly access the attached files. This consists of some basic "header" information for each Result Set created and an "index" that is used to locate all of the different output types that are in the file. Altogether though this information should be very small compared to the total output.

Re: Femap DB size increase w/ attached XDB files

Creator
Creator

Grudy, thanks for the response

 

We get one result set per XDB. Each XDB is 123MB, x600 XDBs. Makes sense that there'd be some metadata stored, but I wasn't expecting gigabytes' worth

Re: Femap DB size increase w/ attached XDB files

Siemens Phenom Siemens Phenom
Siemens Phenom

Thanks.  Sorry to keep asking questions, but we would like to get to the bottom of this and make it better for your situation.

 

Can you give me an idea of ...

  • The number of nodes and elements in your model
  • The typical number of Result Vectors in each Result Set

Just ballpark numbers are fine. We are just trying to get a handle on what parts of the data being kept are causing the problem. Thanks again.

Re: Femap DB size increase w/ attached XDB files

Creator
Creator

Not at all, really appreciate your time and questions. Would be great to be able to work these models in Femap

 

-58,000 nodes and 90,000 elements

-500 output vectors per file

 

Each result file is essentially identical, just different maneuvers

 

Thanks

Re: Femap DB size increase w/ attached XDB files

Siemens Phenom Siemens Phenom
Siemens Phenom

Are you computing any Output?  Linear Combinations? Envelopes? API? etc...

 

If you are, those computed vectors are stored in the FEMAP database since the external attached results files are never touched. 

 

If this is contributing to your issue, with FEMAP v11.2 we introduced a "Results Set Processing" Data Surface that allows you to build Linear Combinations and Envelopes "on the fly" without storing anything in the database.  Don't know yet if that will help you, but just thought I would mention it.

Re: Femap DB size increase w/ attached XDB files

Creator
Creator

No computations, just looking directly at what's in the results file

 

Re: Femap DB size increase w/ attached XDB files

Siemens Valued Contributor Siemens Valued Contributor
Siemens Valued Contributor

Dan26,

 

The increase in the Femap model size if largely due to the xdb index that Femap is builds when attaching to the xdb file.


While all output values are read on the fly as needed some data such as id tables are read once at attach time and stored in the index for fast access. While this approach has provided us with the best performance it has unfortunately caused the index size to grow larger than we desired which becomes an issue when attaching to numerous xdb files.

 

We will look at implementing some improvements that we believe can reduce the index size while maintaining performance.

 

Models that are solved with multiple subcases where results are in a single large xdb file are less likely to exhibit this large growth in the model size.

 

Regards,
Dan